Likely GMP bug

[Marco Bodrato]

Ciao,

Il Sab, 26 Maggio 2018 12:26 pm, Marco Bodrato ha scritto:
> In my opinion, both cases (x >> 32) == x and (x >> 32) == 0 do not give a
> miscomputation,

Thinking twice about this... I fear (x >> 32) == x is a problem.

If V is an odd 32-bits number with a factor in common with 2^33+1, then
entering the loop with vlimb=V>>1, ulimb=2^32=(2^33+1)>>1 will find that
factor != 1. Right?

On the other side, if V is odd and U = kV + 2^32, then GCD(U,V)==1, right?

Ĝis,
m

-- 
http://bodrato.it/